What is your "Oh no my server crashed" plan?

igotacavity · Post by **igotacavity** » Sat Mar 03, 2012 5:18 am

So let's say you have 20 patients scheduled for today, and at 10 AM the server has hardware has issues and goes down. What is your plan to get up and running with minimal impact?

I have thought of the following:

1) Virtualize with VMWare ESX. This is very complicated and will end up costing too much money.
2) Virtualize with VMWare workstation. Less complicated, and if the server crashes, move last backup of VM to another computer which has VMWare workstation. Backups of the VM will get really confusing, and I don't think staff will know what to do if the computer crashes.
3) Get two identical computers. Make one the server. Use Acronis to image the computer every night. If the server crashes, then restore image to the standby computer. Seems like the least complicated and cheapest way.

Things like Windows 2008 and Linux clusters are out of the question. That is an overkill for a practice, and they require a lot of nurturing and making sure everything is working properly. Not to mention the cost.

Hosting offsite is also an option, but the biggest issue is unless you have a really fast connection, X-rays will become slow.

Looking for any ideas to see what others are doing.

Post by **jordansparks** » Sat Mar 03, 2012 8:40 am

Use CrashPlan to make backups to another computer in the office (in addition to offsite and archive). Have MySQL preinstalled on the other computer, but not running. When the server goes down, restore the opendental database to the correct location, and start MySQL. This takes approximately 5 minutes. Simple and cheap.

Jay · Post by **Jay** » Sat Mar 03, 2012 10:12 am

@igotacavity: Jordan's suggestion is spot on. Cloning introduces a new element into the equation, hardware. Why spend on two identical servers and then keep one switched off till the day you restore to it.

@jordan: Crashplan's great if it works. If it doesn't, calling their support is not always fun. They have huge accounts and get irritated with small issues faced by small businesses which ironically stem from the peculiarities of their software. I am also not a fan of port forwarding which they strongly recommend. Anyway, if you must use them, ask for Chad. He Is more helpful than some of the others.

On a different note what works for OD may not work for your X-ray software so check with your vendor. For example XDR (an excellent product) is not installed symmetrically like OD so the server is not the same as a workstation. Cloning wouldn't work though as the installation is tied to the NIC and other hardware of your primary server.

For backing up image folders you can also use Microsoft Sync Toy in Echo mode i.e. left updates right. That way the only thing you have to restore after a crash is the databases which is tiny.

teethdood · Post by **teethdood** » Sat Mar 03, 2012 1:31 pm

You have to think about your backup strategy taking into account whether it is for simple hard drive failure, total server kaput, or theft/Acts of God complete office kaput.

From easiest to hardest to implement:

1) Backup the c:\mysql and all relevant folders to external drive or another computer with MySQL installed ready to go in a heartbeat (Jordan's and Jay's method). Crashplan etc fall under this category as well. This safeguards against total server failure. Five minutes downtime and lots of scrambling around, especially if the IT person (you) is not there. You have to constantly remember to make backups/test your backups. This only works for OpenDental. You would still need to reinstall your X-ray imaging and other programs. Could take way longer than 5 minutes.

2) Hardware RAID. Buy a RAID add-on card and another hard drive. Choose your RAID type. Done, no messing around with Windows. Works for everything including windows 7 Home or basic XP in case your server doesn't run win7pro/ultimate. No downtime, easy to implement. (THIS IS NOT THE SAME AS A BACKUP.)

3) Software RAID-1 the server. Windows 7 Pro/Ultimate only. Stick in another hard drive of similar size and setup RAID-1 mirroring in Disk Management. This safeguards against the most common problem: simple hard drive failure. No downtime whatsoever, easy to implement. (THIS IS NOT THE SAME AS A BACKUP.)

4) Virtualize your server using any number of products out there. Safeguards against total server failure. Safeguards not only your most important files but everything else too. Simply install an instance of the virtual software on your frontdesk computer or buy a new computer and boot up your server image. Five minutes of downtime if total server loss and scrambling around. Lots of time needed figuring it out initially.

Personally I have #3 and #4 implemented (I actually use Linux RAID but the idea is the same). For your everyday office, I would suggest going with #1. If you're a tinkerer, then do #2 or #3 and/or #4 .

jimgaas · Post by **jimgaas** » Mon Mar 05, 2012 11:44 am

You are sure right to be thinking about this before it happens!

It is interestesting that you said "your server goes down." How it goes down makes all the difference in the world. If the work horse of your computer goes bad, namely the disk drive, you are golden if you have a clone.
If your server goes down due to a bad motherboard, or something other than the disk drive that is a different story. A clone or the original working drive will most likely not work in another computer. You might get very lucky and have a computer with a very similar architecture and only have to revalidate through Microsoft. Or every Windows product around may tell you your drive is junk and has to be reformatted. This happened to me. My motherboard went bad and the only thing I could make read my
disk was Linux. Linux saw it right away and all the data on it. This was a windows disk and no windows operating system could not see anything on the disk. If it is your plan to recover this way, you should try it out in advance of the day of disaster.

Here is our plan for disaster:

1. Clone the disk sequentially. This saves everything including all network settings, all data...EVERYTHING. We had an update that trashed our network and it happened such that we wound up cloning the bad stuff so our 1 day old clone was also junk. We had to go back two days. So, clone everyday and make multiple clones, I would recommend everyday for a week. My server has a bay on the front of the computer to put a sata clone drive into and out of the computer.
2. I like your idea about virtualization but it can be free. I am making my front desk person set up a virtual machine in Virtualbox. She can configure it to her liking. I told her to make snapshots and to occasionally export the appliance. If I change computers, I will only have to import the appliance on which she is used to working. I also use this on work stations for the same reason. For example, the old black Schick hub will only work with 32 bit windows products. If you have sufficient ram, you can run a 32 virtual machine on a 64 bit computer and use your old Schick hubs without having to junk them forcing you to go with the new white hubs that support 64 bit computer operating systems. In addition, you can have an appliance that is on the desktop or even on a CD ROM so that if you get a virus you just load a new VM. If you get a new computer that is not configured for xrays, for example, just load a VM and off you go.
3. Make very long term sequential backups of just your data. I use a batch file that writes a new date to a folder it creates and I just put the same hard disk in until I need to delete some of the old files. This helped me to figure out if some missing xrays were never taken or if they were lost somehow. If your backup leaves the office, it needs to be encrypted. Remember to figure out what your DATA consists of. It is your A-Z folder, your xrays, your MySQL database, and in my case an MS SQL database used for my Schick xrays. Everyone will have different stuff to back up.

Raids arrays are great to guard against drive failure but they will not help you when someone accidentally deletes a file as happened to me. I had to restore from a file that was not on the server.

This is an interesting and helpful thread.

joshuab · Post by **joshuab** » Mon Mar 05, 2012 12:36 pm

We're syncing our images folder to an additional computer, besides backup. I see OD has an option for sourcing from two - not sure if it writes to one or both or how this works. Intend to explore adding this soon for redundancy.

I just setup mysqldump to export the entire mysql db to another computer every 20 minutes. Haven't used CrashPlan - does it sync the data directory live? If no, does mysqldump get EVERYTHING required to restore from? If so, could restore this automatically to a another computer (workstation) running mysql and it would be ready/live all the time.

We use Macirum (works better IMO and much cheaper than Acronis - have tried both) for nightly image backups.

As Jay mentioned other applications are another problem. We're running (CRAPPY) Schick sensors/app due to previous (costly) investment. It runs on MSSQL and whole other beast I have to figure out how to restore from, besides the image backups.

jimgaas · Post by **jimgaas** » Mon Mar 05, 2012 1:17 pm

To joshuab:

You can rebuild the database if you backup the image directory for Schick.
The directions are on the disk that came with CDRDicom.
I would love it if Jordan built a PACS server for everyone's Dicom images.
Maybe it is intended to be part of the A-Z folder at some point since that seems
to be mostly empty. If we could transfer our images into the MySQL database structure,
that would make backing up a lot easier and get rid of the bridges, hopefully.

Check out MicroDicom. If you point this free viewer at the images directory, it will
load all the images into the viewer. You can zoom and look at the images as a set,
like a full mouth xray, etc. When I pointed the viewer's menu at "Find Dicom Images"
it loaded 30,000 images into the viewer for all the patients with digital xrays. That
could save you in a pinch if your server crashed. You just need those images and you
are good to go.

Jay · Post by **Jay** » Tue Mar 06, 2012 8:40 am

jimgaas wrote:You are sure right to be thinking about this before it happens!
It is interestesting that you said "your server goes down." How it goes down makes all the difference in the world. If the work horse of your computer goes bad, namely the disk drive, you are golden if you have a clone.
.
.
.
This is an interesting and helpful thread.

This is a great thread especially because of your posts. I liked the bit about MicroDicom.

Re: disk cloning, I am all for it if you use it to restore the computer that was cloned but not across machines because then you have to match hardware which is a tricky or expensive or both. Across machines I prefer to have a lesser machine (no RAID array, no Dual/Quad core processor, no dual NIC mother board, no USB3/e-Sata card, no hot swap-able drives or power supplies) that is running with MySQL installed and OpendentalImages already copied over using Sync Toy in Echo mode plus Windows Scheduler. Then if the main server goes down, you just take the latest database dump and restore OD to the lesser machine and point all other workstations to it. For XRAY software there is also the additional step of configuring the server on the backup server before you restore the databse but the image files themselves can again be SyncToyed!

@joshuab: For those not keen to fool around with mySQLdump there is a software called dbForge MySQL backup tool. For $99 you get a GUI and together with Windows Scheduler you are good to go. http://www.devart.com/dbforge/mysql/stu ... ackup.html

bpcomp · Post by **bpcomp** » Tue Mar 06, 2012 2:13 pm

Just to add to the mix is a commercial disk cloning product by Symantec called Backup Exec System Recovery. It will do a full backup of the running machine with hot database like many other products. What sets this software apart is the ability to restore to dissimilar hardware. I used to work in the IT industry and there were several times where a server would fail and we would bring in different hardware and restore to it having the system back up and running within usually 8 hours. The client used the "loaner" server until their new server arrived. Also because of the way it backs up your server you are able to restore the backup to a virtual machine for testing.

This is not an easy choice because of the price, but for anyone who is interested...

http://www.symantec.com/system-recovery-server-edition

(I do not work for Symantec, I only bring it up because of the previous comments discussing the inability to restore to dissimilar hardware)

Justin Shafer · Post by **Justin Shafer** » Thu Mar 08, 2012 7:51 am

Jordans method is good. Just to be sure that you update mysql-server on your workstation from time to time. Test to make sure everything works. Its easier to rename the computer name to what your server name was, unless your using a domain. Its like a homemade disaster recovery plan. Not as nice as ESX, but not as expensive. ESX is better for quickbooks, etc.

shvercer · Post by **shvercer** » Mon Mar 12, 2012 12:35 am

Jay wrote: @joshuab: For those not keen to fool around with mySQLdump there is a software called dbForge MySQL backup tool. For $99 you get a GUI and together with Windows Scheduler you are good to go. http://www.devart.com/dbforge/mysql/stu ... ackup.html

I wanted to know if anybody uses this dbForge product. I've looked at their website and there doesn't seem to be any contact info other than e-mail addresses and a form that you can fill out. Doing a WHOIS lookup on their web address doesn't lead anywhere. I'm a little bit wary given what the article in this forum post states:
viewtopic.php?f=2&t=4306&p=17766#p17766

Since there is no person, address or telephone number on the website it seems somewhat odd.

Jay · Post by **Jay** » Mon Mar 12, 2012 12:36 pm

While you are right to be concerned I cannot guarantee anything and you must make the decision yourself. It does seems highly unlikely that a company which publicly lists as clients several Fortune 500 companies would be up to no good. But I have no connection whatsoever with this company so I cannot say any more.

Customers: http://www.devart.com/company/customers.html

drmaximus · Post by **drmaximus** » Tue Mar 13, 2012 5:31 pm

For everyone using crashplan do you have to STOP the mysql service in order to back it up or is crashplan able to backup up all the files while the service is running? I am thinking about switching from MOZY. It seemed Mozy was able to backup my database while the MYSQL service was still running.

Jay · Post by **Jay** » Tue Mar 13, 2012 6:54 pm

Crashplan can do live backups but it has a few peculiarities:

1. The recommended installation involves opening and forwarding a port on your router. It might work anyway but if it doesn't then tech support will urge you to do it.

2. It doesnt mix well with dual NIC systems which confuse it.

3. The documentation is Mac centric, not that there is anything wrong with that.

4. Tech support quality is competent but does not go that extra mile. Basically they have huge accounts and get irritated if small fry like us present problems that require unique solutions. They will say something like: "You cannot expect me to figure out how you should set up your network/NIC/OS/router/antivirus (pick one). When you have finished settin up your machine properly, please try again." If you say that everything else works perfectly on your PC they'll repeat the same line.

So if Crashplan works for you, great. If not, you might find yourself on your own.

Post by **jordansparks** » Wed Mar 14, 2012 7:23 pm

Jay wrote:"You cannot expect me to figure out how you should set up your network/NIC/OS/router/antivirus (pick one)."

We should try that line with our customers. It would sure save us a lot of time.

Jay · Post by **Jay** » Thu Mar 15, 2012 7:03 am

jordansparks wrote:
Jay wrote:"You cannot expect me to figure out how you should set up your network/NIC/OS/router/antivirus (pick one)."
We should try that line with our customers. It would sure save us a lot of time.

We are all spoiled by OD. Even XDR which has a great (well deserved) reputation for customer support pales in comparison.

jimgaas · Post by **jimgaas** » Fri Mar 16, 2012 3:35 pm

I think the thing that keeps getting overlooked in the backup discussions is the network settings are not backed up with file by file backup methods like Crashplan appears to me to be. Jordan is right on when he says backing up your data will keep you from going out of business. Everyone here is looking to get back up on your feet quickly in a disaster. The first question was how do I recover to save a day full of patients as quickly as I can. If all you had to worry about is Opendental, then everything said holds true and your data alone will get you up and running very fast. But where you may get a rude surprise is your xray stuff. I can tell you that configuring my xray database stuff is very unfriendly. If I lost my server and I couldn't immediately replace it with a clone that was identically set network wise, I would lose at least one day of taking xrays. I would suggest that anyone who is serious about backup, should take an office computer that is not your server, put all your current data on it, unplug your server and plug in the replacement computer as your new server and see what happens. This is exactly what will happen when your server goes down and you get your data back from Crashplan or where ever else your have your data. Your will have all your current data on a computer that is not your server. This computer will not function like your server did. It would if only Opendental was involved, but as I mention, see what happens with your xrays. In the past I have had a legal legitimate version of Windows force me to revalidate because too many peripherals were changed in the computer even though the CPU and motherboard where the same. So you have Windows working against you if you want to make your server hard drive work in a different computer. I recently bought a new computer that is the exact model I use for my server and to my amazement the hard drive clone transferred to the new computer and it worked perfectly. I have never had this work before and maybe Microsoft has loosened up their security due to pressure from people who know the importance of getting a backup to work on different hardware. You have always been able to do this in Linux. So the bottom line on quick recovery from losing your server is this, clone your server hard disk drive (do not use the original drive in the server in case you mess it up), take the CLONE of the server hard drive and try it in another computer. If it appears to run, then unplug your server, plug in the replacement clone, and check the workstations. Particularly, try taking xrays and see if the get stored properly and can be retrieved as you expect.

stjames70 · Post by **stjames70** » Fri Mar 16, 2012 10:37 pm

Right. That is why the original post was correct. Copying a virtual machine is not complicated and VMs are truly hardware agnostic as long as said hardware runs fast enough.
Between OD, Dexis, Quicken, Excel and TimeClock Manager, that pretty much comprises my core apps, so hardware overhead is really minimal.
I have two MacPros: one at home and one at the office. They are clones of each other and they are synched manually by copying the backup files of OD, Quicken, Dexis, Excel, and TimeClock Manager every night to ensure the fidelity of the backups. We would lose at most, one day of data in case our office server just happened to fail.
I would bring the second MacPro to the office, plug it in the wall, and since it is a clone of the original, all the other computers would not be able to tell that it was even replaced.

RickInHouston · Post by **RickInHouston** » Sat Apr 07, 2012 8:20 am

I’m currently faced with this issue. I have a few questions.

I’m running a Windows XP dedicated server for the last 8 years. I’ve been running Soft Dent for the first 5 years and Open Dental for the last 3 years, approximately. My server has stopped running about three times the last two weeks and suspect the power supply is going out. I’m building the new server myself. I have 8 computers connected to this server.

Once the new computer is up and running and have a chance to shut down the office:

1. Rename old server, named SERVER, to something else. Name new server this and add it to the network.
2. Download Open Dental trial and install it on the new server.
3. Copy Open Dental and mysql folder from old computer to new server (do I have this right?) folders of the same name.
4. Copy Apteryx folder from old server to new server (Apteryx says that’s all that is necessary to have their imaging program up and running again. I’m not sure I believe this but will give it a try).
5. Does anyone know if I can copy X-charge and Tesia folders to new server from the old server, as well, or do they have to be installed?
I’ve never ‘installed’ a program just by copying the folder over to a different program and didn’t even know this works. Does it?

Please give me any heads up information on this, if you have any. I’d much appreciate it,

Rick Maloch
Houston, TX

Post by **jordansparks** » Sat Apr 07, 2012 5:42 pm

3. Almost. You'll want the OpenDentImages folder and the mysql folder.
3b. Open port 3306
3c. Share OpenDentImages folder.

RickInHouston · Post by **RickInHouston** » Sat Apr 07, 2012 6:14 pm

Thank you.

bpcomp · Post by **bpcomp** » Mon Apr 09, 2012 1:48 pm

5. No you can not just copy folders and have the programs run in most cases. Installation of the program and copying data is usually OK. Some programs require you to export data in a special way. If you just copy a folder that has a program in it, it doesn't insert registry keys or register DLL's or anything else that happens during an install. Unless you have a program designed to run as portable or standalone... go with a full install and try to copy data.

Post by **jordansparks** » Mon Apr 09, 2012 3:04 pm

5. Right. Both XCharge and Tesia have programs that get installed as well as data folders that get moved.

RickInHouston · Post by **RickInHouston** » Wed Apr 11, 2012 7:18 am

Here’s my actual real-world events which took place yesterday April 10. 2012:

My old Dell Optiplex ‘server’ running Windows XP for eight computers was shutting down here and there for the past two weeks. It had been running pretty much perfectly for 8 years. I felt it was time for an overhaul.

I bought all the parts from New Egg for a new computer. Improvements I wanted to make were: larger metal case for heat dissipation, non-proprietary parts so switching them out could be easier. I increased the amount and size of the fans to push air through the case. I updated OS to Windows 7 Ultimate.

I put the new computer together and booted it up. I installed a fresh copy of Windows 7 on new machine, installed the motherboard drivers, installed all Microsoft updates to the OS, and installed Microsoft Office on new server.

For Open Dental: I installed the trial version of Open Dental. I installed .net 4.0 on the new machine. I copied mysql and OpenDentalImages onto the new machine. I shared the OpenDentalImages folder. I openned the appropriate port (you can type in the help box (how to open a port to get an easy walk-through of this).

For Apteryx (if anyone is using it): you only need to copy Apteryx folder and all its contents to the C: drive of the server. It does not have to be installed. Then you must call them and re-register your licenses. I shared the Apteryx folder.

Installed Karen’s Replicator and used old settings to set up new settings to backup mysql, OpenDentImages, Apteryx and a folder of files I use called Workstation Info (patient education, forms for the office to print, etc that can be accessed from any workstation) nightly. I don’t do incremental backups, only full backups after each day of work, not weekends.

I copied mysql, OpenDentImages, and Apteryx to a removable drive. The office must now go off-line.

Here’s where it went wonky. I began with DocAide, then to Softdent and on to Open Dental. My workstations each had a mapped drive to the server on them. I thought if I simply renamed my new machine to the name of the original server and took the original server and renamed it (ServerOld) it would be an easy switch. The server could not be accessed this way. I went to each workstation and deleted (disconnected) the mapped drive to the server. Open Dental support (go Amy!) got each workstation to simply connect directly to the new server in network connections without using a mapped drive.

Tesia works without any additional support issues. I think Open Dental did make sure the path (Setup > Program Links > Tesia) the was correct.

X-Charge needed a phone call and remote connection to reinstall on the server and update each workstation where a card swipe is installed (Front desk two computers). Having my old server still running on my network made it easy for them to copy pertinent files to have my credit card history placed in the newly installed version so I can access old information easily (great to have available).

In Windows 7 Ultimate: change power management settings so hard drive never sleeps or backups will not happen at night. When I was ready to put the new server into my network and transfer the most current copies of all folders (taking the office offline) it was a two hour process because of the hiccup of getting workstations to talk with each other. Once Open Dental was up and running the office could start functioning with just that. Adding my digital imaging program took about 30 minutes to get up and running, X charge took a good 45 minutes to get working. That is my time frame and I’m not very IT oriented. Others here on the boards can do this much faster. Having your workstations connected this new way would probably allow a simple switch of server my switching names like I previously thought might be possible. This would have made things go much faster and much smoother.

Now is probably a good time to perform a Windows recovery copy (or whatever it’s called) of the hard drive since my new drive is fresh, clean and pristine. In the event of a hard drive failure I can install a new hard drive, recover snapshot from this recovery, copy over current Apteryx, mysql and Open Dental folder and be up and running with a hard drive failure. This is something many should do RIGHT NOW to prepare them for a hard drive failure if they don’t have some sort of RAID or mirrored drive happening now.

How does this all sound?

bpcomp · Post by **bpcomp** » Wed Apr 11, 2012 8:14 am

If you are up and running then that is great. I'm a little fuzzy on your backup solution. Do you stop the MySql server before running Karen Replicator and do you ever test the backup that it makes? If you don't stop the MySql service every night before running the backup then you might want to consider looking at Duplicati. It does allow you to copy a hot database and restore which I have tested. What all that means is that you don't have to babysit the backup every day after work. It also allows you to upload to an online storage (encrypted of course).

RickInHouston · Post by **RickInHouston** » Wed Apr 11, 2012 12:35 pm

Thanks for the heads up, bpcomp! I have never stopped the mysql program (and don't even know how to do that) to back it up). For some reason how I got back up and running, without stopping the service, worked! I'll have to look into another plan as you have suggested to solve the problem you brought to my attention.

B.Thomas · Post by **B.Thomas** » Thu Apr 12, 2012 4:31 pm

Well, it happened to me yesterday and it wasn't fun. Fortunately the server came back to life after I used Windows System Restore. I think a recent Windows update caused the problem. It just kept rebooting. But of course it happened during the busiest part of our day causing us to create makeshift charts to take notes on as I ran back and forth to try to restore a backup on another computer. Very Stressful!

I want something I can switch to "on the fly" at a reasonable cost so that all the client computers running OpenDental will point to the backup server quickly and easily. I envision something like this called a "failover cluster"

http://docs.intersystems.com/cache20102 ... lover_cold

I'm assuming I would need Windows Server and 2 identical (or close to identical) servers. I've never worked with Windows Server so I don't know if I would need additional software to handle the switch-over in an event of a failure.

RickInHouston · Post by **RickInHouston** » Fri Apr 13, 2012 5:35 pm

I had two good days with the new server. Went to make a recovery copy of the main drive and it stopped 75% of the way. It said to run chkdsk on the drive. Drive showed errors and it failed to boot from that point forward. Took the drive out to read it with another machine and it couldn't be read. Womp, womp, womp. I was SO CLOSE to making a copy of it, damnit.

Now back to reinstalling everything on a newly purchased hard drive. Oh, well! At least I know what steps I have to go through!

mikemee · Post by **mikemee** » Wed May 09, 2012 8:37 am

As "an IT guy" who helped my neighbor find OD and helped them through the install & subsequent running (>4 yrs now iirc) of their office, this is how they handle the "what if the server fails" scenario for their office, fwiw:

1) a good quality rackmount server (e.g. Dell) with a 'server grade' hard disk (it was enough to specify SCSI at that time, but buying a 'real' server is 90% of the battle). If I was buying one now, I'd go with a simple dual drive RAID-1 configuration, probably, though in principle I'm not a big fan of RAID as it usually adds more complications and failure points than it solves. The server is not a place to skimp. Although I'm a hacker who has built computers many times and enjoys mixing and matching components for the best price / performance, hardware has become so cheap, and downtime so expensive, that I now shy away from doing this for friends/clients. Dell provides great warranties, with same day onsite service and/or overnight replacement parts. The server hardware specs tend to stay stable for years - which means you can often simply remove a hard disk from one machine and put it in a replacement box and 'it just works' with the same drivers etc. And, you also avoid lots of subtle DIY hardware problems that otherwise manifest as blue screen / server reboots because, dang it, the server is designed to run 24x7 and *never* crash.

This server wasn't a popular choice at the time, as it easily cost 2x the equivalent desktop and the disk drive was much smaller for the same price as the equivalent desktop drive. It was also inconveniently big/awkwardly shaped and noisy. However, touch wood, that server is still running, has never hiccuped and we're only now talking about replacing it because "4 yrs is a long time for any computer" (that, and a 2nd internal drive we added 2 yrs later seems to be starting to fail, more on that below).

2) have a nominated machine as the "if the server fails" machine. In this case its the dr's machine which is only used for email/web surfing. This is also the machine that is used for testing major upgrades of OD and other similar tasks. I.e., it somewhat regularly (~3-4 times/yr) has a backup restored to it and has another machine running OD "pointed" to it temporarily.

3) have a set of 3 USB hard drives that act as weekly rotating backups that are taken offsite. A full backup is done automatically each weekend and the drive is swapped out sometime during the week in preparation for the next week. Why 3? Nominally, this is to catch the "oh crud, this got deleted 10 days ago, but we only just noticed - good thing we have a backup from 2 weeks ago as we now know that last week's backup won't have it". Realistically, its also a great comfort to have the backup of the backup of the backup. E.g. as you plug in the USB backup drive to retrieve a file that got deleted accidentally, the drive falls off the desk onto the floor and makes a horrible sound as it lands. More drives than 3, and it starts to get confusing, in my experience. Oh, and don't buy 3 drives at the same time and place for this backup set: hard drives tend to fail in batches (i.e. similar serial #s). Add a new drive every 12 months or so and retire the oldest.

4) another USB drive is 'permanently' attached to the server for nightly backups. This, together with the weekly backup, can be used to bring up a new system to match the end of business the day before. This drive should be exchanged for one of the nightly backup drives every few months to distribute the wear.

Overall, we've become big fans of USB attached drives. Surprisingly, their performance with general OD use, including X-Rays has been excellent. We learned this the hard way when the added drive (with all the images on it) in the server started to fail, we plugged in the most recent USB backup drive temporarily to keep the office running until it was convenient to replace the drive -- and, if anything, performance improved. Personally, I suspect this is because the other drive had been failing for a while and was slow from retries, but ... in any case, it suggests another backup / recovery model. A USB drive can easily be moved to the "if the server fails" machine and be ready to go immediately. Indeed, any of the backup USB drives can be used that way: plug in the drive, share out the image directory from the backup dir, adjust the OD clients (or adjust your router DNS to point to the new machine) and you're done. (This assumes that your backups are simply copies of the original files, rather than some complicated encrypted compressed backup). That said, I wouldn't recommend running MySQL on a USB drive (though it might work just fine!).

As a further fwiw addendum, they're running CentOS (Linux) on their server. This saved a bunch of money on a Windows Server license, removed hassles with updates and has been super stable. It has various downsides including no direct OD support for things like upgrading MySQL or setting up shares. Realistically it only works for their scenario because the dr's spouse has decent *nix skills and can do the usual things like mount drives, run/tweak rsync scripts for backups, and adjust Samba shares. I'm not recommending this as a solution for most people, but I mention it in passing in case someone with *nix skills is curious. Certainly for them, it has resulted in a *very* stable server setup. (Note that the "if server fails" machine is regular Windows - there's no problem mixing and matching here).

Whew, a bit longer than I'd intended. Hope its useful.

B.Thomas · Post by **B.Thomas** » Wed May 09, 2012 1:03 pm

mikemee wrote:As "an IT guy" who helped my neighbor find OD and helped them through the install & subsequent running (>4 yrs now iirc) of their office, this is how they handle the "what if the server fails" scenario for their office, fwiw: Whew, a bit longer than I'd intended. Hope its useful.

Very useful info. Thanks! I am still intrigued by the "Failover Cluster" setup. I have been experimenting at home with a trial version of Windows Server. It's surprisingly not that difficult to set up since it looks like Windows 7. However, the licensing costs is a negative. Do you know of any other Open Sourced Server operating systems?

Also what software would you suggest for running "Thin Clients"? This has a lot of appeal to me because it means I only have to install and update programs once instead of on 9 different computers.

mikemee · Post by **mikemee** » Wed May 09, 2012 9:15 pm

Its easy to start sliding down the slope of "hot spares", "failover clusters", "server attached storage" ... etc., etc. There are many opportunities to geek out here, way beyond what the requirements will justify. Standing back, its worth taking a look at the cost of a server failure at a bad time for your particular office.

E.g. I suspect a small one-dr practice can survive with a nominated "if the server fails" machine for a day if need be -- e.g. while waiting for "the tech person" to come by, restore a backup and get everything online again.

However, a large many-dr practice will be seriously hurting if the server goes offline for even a few minutes, and this would justify some sort of dynamic failover. (By the same logic, you probably want backup internet connections for voip-phone systems etc.) Even here, though, there are many ways to do this, and having a standby system that is up to date as of last night might be a lot cheaper and easier to keep configured than a complex failover cluster solution.

Thin clients are interesting, but again add another layer of complexity and also introduce a single point of failure (the server hosting the clients). They might also be more expensive. Again, possibly very much worth the effort in a large office with 10-15+ client computers, but not with a handful.

One random thought is cloud hosted databases. With a decent internet connection, it might be possible to have your MySQL server hosted at Amazon. Amazon offers a plethora of nice management solutions, including hot backups, redundancy etc., etc. I doubt that xrays would work well this way because of the huge files, but having a local NAS (Network Attached Storage http://en.wikipedia.org/wiki/Network-attached_storage) box with a continuous network backup could be v nice. I'm not completely updated on the latest OD requirements, but I think that in theory, with a cloud database server and a local NAS, you can do away with the need for a local server completely, and focus your effort on keeping the internet running (e.g. with dual internet feeds and a router that supports transparent failover). (Perhaps cloud database services have been discussed and benchmarked here already, I'll take a quick look and update this post if so).

mikemee · Post by **mikemee** » Wed May 09, 2012 9:21 pm

B.Thomas wrote:Do you know of any other Open Sourced Server operating systems?

Sorry, missed this part. Most open sourced servers are variants of *nix (Linux, Unix). Personally, I would use whatever the hardware vendor was offering, as it will be something that they've heavily tested the drivers and overall configuration (e.g. Dell has CentOS (aka RedHat) and possibly Ubuntu - I haven't checked in a while). The more mainstream the OS, the easier it is to find online guides and also local geek support for. That, of course, is the other big factor. If your local trusted tech person knows X, then best to go with X! (Nothing worse than a client who insists on their own solution, overriding your recommendations, and then requires you to fix it when something goes wrong. Better to find another tech person if you won't accept their solution imnsho.)

Jay · Post by **Jay** » Thu May 10, 2012 7:16 am

@mikemee: Very detailed and useful post. Can you elaborate on two points that'll make it even more useful:

1. What software do you use to make backups?

2. How do you backup open databases?

3. Do you do offsite backups over the internet? If so, what service do you use? If not, why not?

mikemee · Post by **mikemee** » Thu May 10, 2012 8:01 am

1. What software do you use to make backups?

As its a Linux box, we use the standard rsync utility. It simply does an efficient copy from the source to the backup drive. Drive space is cheap and rsync is fast, so there's no attempt to compress or make "diffs".

2. How do you backup open databases?

Using the Mysql backup utility. This is run by a script at night when OD isn't being used.

3. Do you do offsite backups over the internet? If so, what service do you use? If not, why not?

No backups over the internet. My client says its all working just as it is and they don't mind changing the drives regularly. I've become a recent fan of online backups and have suggested it, but there are some complications (e.g. they use Linux and most of the convenient automated systems don't support that). My favourite for Windows/Macs is http://www.backblaze.com/.

wjstarck · Post by **wjstarck** » Thu May 10, 2012 8:04 am

Really thoughtful and useful discussion.

Just wanted to add a footnote: OD currently does not support multiple instances of the same plugin, so if you are planning to use plugins at any point, thin clients will not work. Not a big deal right now as there are not many plugin authors, but might become an issue down the road as OD userbase grows.

Ralph McElmurry · Post by **Ralph McElmurry** » Thu May 10, 2012 9:04 am

We run two identical rack servers. Each is RAID 5 with 6 hard drives. The production server backs up to the other nightly and also to a muscular work station elsewhere in the office. Both rack servers are Linux running character based only with MySQL. The images files are shared to workstations with Samba. The Vixwin xray data has a similar scheme. All the data gets copied to a rotating usb hard drive and tested on a server at home. If the main server goes down, it is a simple matter to simply logon to the backup. There is little danger in inadvertently logging on to the wrong one since the staff does not know its name and it does not advertise on the database chooser. We have two separate practices on the system and about 18 workstations

mikemee · Post by **mikemee** » Thu May 10, 2012 10:00 am

Ralph McElmurry wrote:We run two identical rack servers ... <snip>

Very nice setup! Sounds well thought out, and, tested! (A common failing of backup processes is that people never sit down and actually walk through a full recovery - its stressful to be doing it for the first time when the pressure is on

)

bcpayne · Post by **bcpayne** » Sun May 13, 2012 5:26 am

my current setup has a speedy server running windows 7 professional. On that I have virtual box running with a virtual ubuntu server which is my file server and mysql server. my virtual server and drives are stored on two 1 Tb drives in a windows mirror (software raid 1). If my server or one of the drives goes down I can remove 1 or both of the drives, insert them into another window 7 pro computer, start the virtual machine in virtualbox and be up and running. none of the workstations need to be redirected because it looks like the exact same server on the network.
the virtual ubuntu server is running crashplan pro, so it is constantly being backed up online. we use that solely for all file storage, so everything is backed up for our office with only one computer account at $7.50 per month unlimited.
I haven't tested a recovery from crashplan. A-Z should be fine, but I have a feeling that, depending on whether mysql was in use at the time of backup, it may not be in a reliable state to recover from. Anybody else restore from crashplan?
I still want to setup a backup from the virtual server to another computer in the office every 10 minutes with a scheduled mysqldump or a commercial mysql backup program. that should close the hole of risk of frying both raid drives at once and a day old crashplan backup with hours of work lost.

Justin Shafer · Post by **Justin Shafer** » Sun Jul 01, 2012 7:27 am

Why not go for a mysql-cluster? Then all you have to worry about are the files on the fileshares... I am going to do this for fun instead of mysql dumps...

I have done this with linux with heartbeat and drbd.. But I want a windows solution that runs a cluster... ie when a server goes down the other server has all the data that the downed server has.

asynchronous replication. mysql does have their cluster software, but I remember something about primary keys and the database....

Regarding the files on the file share...Linux did this well with drbd, smb, and heartbeat. It is expensive to do asynchronous replication of fileshares with windows. Free with linux. For that you just may want to use something that will send over changed files every x amount of minutes.. Dont use DFS (Distributed File System) which is an option in Windows Server.. not to be used for databases and that might be your x-ray database.

Fail-over with heartbeat was great. Clients never knew the server went down. Didnt have to change the primary keys either.

Post by **jordansparks** » Mon Jul 02, 2012 10:59 am

Open Dental will NOT work with mysql cluster. Table rows are too long.

Justin Shafer · Post by **Justin Shafer** » Sat Jul 07, 2012 8:15 am

Ahh.... then the DRBD/Heartbeat Solution is still going to be the best cluster method other then Microsoft Cluster and that is expensive! I tested it with San Melody and Steel-Eye.. It worked, but was pricey.

DRBD/Heartbeat had a faster failover too.

This is good because it uses the standard mysql app and database. The database sits in the DRBD partition and can be used by mysql on either server... Heartbeat controls the mounting, and choosing which server is running mysql and which one is not.

cheroxy · Post by **cheroxy** » Thu Aug 16, 2012 12:37 pm

Here is my plan as outlined today for my office. Just another version of the simple "move server to another workstation" in easy to follow steps

Step-by-Step Restore

1. Go to computer that will be the temporary server
2. Note what version of Open Dental the office is currently running (12.2.16.0)
3. Download trial
4. Install as “New Server,” but uncheck the Open Dental Program install, follow the instructions and do “Typical” MySQL install not custom or complete.
5. If last backup was on a different computer than the one you are using as the temporary server, copy the last backup to this temporary server.
6. I'm not sure if it is really needed, but rename the backup version of the folder “opendentimages” to “OpenDentImages”
7. Replace the following folders with the backup versions by pasting the backups over the existing ones (use the move and replace option):
“opendental” this folder is found in C:\mysql\data
“opendentimages” this folder is found in C drive, it is not under any other folders
8. Share these folders by right clicking on them, select “properties” then click on the “Sharing” tab. Click “Advanced Sharing” then select the “Share this folder” box. Don't change name of folder.
9. Open Open Dental on the temporary server and change database from “SERVER” to “localhost” and see if the update is working. If you can't solve any problems from here on you can at least use this computer to enter all the data in until the others can be connected to it.
10. Note the name of the computer you are using as the temporary server (today is PRIVATEOFFICE)
11. Open port 3306:
Open Windows Firewall by clicking the Start button, and then clicking Control Panel. In the search box, type firewall, and then click Windows Firewall.
In the left pane, click Advanced settings. If you're prompted for an administrator password or confirmation, type the password or provide confirmation.
In the Windows Firewall with Advanced Security dialog box, in the left pane, click Inbound Rules, and then, in the right pane, click New Rule.
Follow the instructions in the New Inbound Rule wizard:
Select port and click next
Select “Specific local ports:” and enter 3306 and click next
Choose “Allow the connection” and click next, next, and name the rule “OD port”
12. Go to each workstation, log on to Open Dental with Administrator access and click File, Choose Databse
13. In the Server Name box change it from “SERVER” to whatever your temporary server's computer name is as noted in step 10 above
14. Click ok and that should be it

(Print these instructions and keep taped to server)

tdong · Post by **tdong** » Fri Aug 17, 2012 7:35 am

Two identical servers mirror by the software call Double-Take (this is bytes level mirror). They check for fail over using ip ping and email me if there is any problem. Two option for the dental office. One is to use OD database and connect to server two. One problem with Kodak xray is that they only allow one server to run; therefore I have to start the service manually (run the .bat script).. Second is to call me (if I didn't check my emails) and I will do the fail over which take around 5min via logmein. This software also allow me to write scripts to start other windows services. After the fail over the mirror server has the ip of the main server. There is an automatic fail over option which you can set but I rather have control over the fail over. After I fix the main server I can do the fail back which will copy all the changes from mirror server back to the main and set the main as main.

also have a backup copy of OD database write on a share folder of a laptop every two hours. The laptop can be a stand alone server or the assistant can carry the laptop from op to op ...heheh worst case I guess

. and the daily backup to external hard drive.

I must admit dell server is amazing. We run them for the last 6 years without any hickup. I only update and reboot the server once every 2 to 3 months. I only use raid 1 since the database is so small. OS is raid 1 (2 76gig) Data raid 1 (2 500g). at my work place raid 5 failed at least once every 2 years. Raid 1 is the most stable Raid setup

Justin Shafer · Post by **Justin Shafer** » Sat Sep 15, 2012 11:36 pm

http://community.spiceworks.com/topic/1 ... nvironment

I thought about messing with DRBD again.. and I think ProxMox is getting more mature.. mature enough to test. ProxMox is basically a gui to DRBD.

http://pve.proxmox.com/wiki/Cluster_con ... on_(Video)

You use 2 (3 is better for a quorum) to create 2 servers that are always in sync. Its like VMWare VSphere.. but free.

Justin Shafer · Post by **Justin Shafer** » Sun Sep 16, 2012 3:33 pm

Oh yeah, and it uses File System Replication that is truly synchronous.. No SAN required.

stjames70 · Post by **stjames70** » Mon Oct 22, 2012 5:50 pm

Hey, I thought I would just put in my two cents. If you look at some of my other posts, you will see that I am a big proponent of VMs. I have a good friend that works for Western Union, and for the most part they run strictly VMs throughout their enterprise. And the reason is simple -- data loss can be minimized, it is fairly fast to restart a VM, and VMs are mostly hardware agnostic so you can restore a VM on almost any hardware. We run VM Ware Fusion in our machines. The computer which holds our database, our MacPro, has four drives. I use one drive to hold OS X and VM Ware, another drive to store the VM itself, and the remaining two drives are simply mirrors of the OS X and the VM. We use Carbon Copy Cloner which very efficiently copies those drives nightly. Just in case, I also back it those drives to two separate external drives. I have booted the computer from the cloned copies and we do have data loss of about 24 hours since the data is only copied once a day, but you know, if that is the worst case scenario where we lose one day of data, I think I can live with that. Now, if more than just the drives fail, we do have a second Mac Pro stored off site which can boot up right away after the cloned drives are inserted into it.
Your hard drives will fail. That is a given. How much time you want to spend rebuilding your server or your workstations is up to you, but with VM clones, you can be up and running pretty quick. Large institutions cannot be out of service for too long. They use VMs because of the reasons I mentioned above. Scale this down, and you can see why you probably should be running VMs. You don't have to resort to fancy software or hardware schemes, and you can pretty much boot up your VM in any piece of hardware that is fast enough and be up and running. You don't have to reload your software, you don't have to open ports, you don't have to reinstall SQL, you don't even have to redirect your workstations to your new server instance since it is a clone of your previous server. It takes me two hours to reboot a new server from a cloned copy. How long does it take you to restart?

Justin Shafer · Post by **Justin Shafer** » Sat Nov 10, 2012 7:58 am

I will say that non enterprise drives really have hit the toilet lately... Seagate and Western Digital have been a real gamble.. Even with workstations that have Raid.... I setup a home pc with raid 1 for a doc (dell) and it had Intel Storage Manager loaded in the system tray... Apparently one of the barracuda's started to go out... and Intel started to warn the doctor about the volume not being redundant.. He ignored it for awhile.. Then the other Barracuda started to give way...

After Windows was bluescreening he finally called.. I got all the data off... luckily... Because he unplugged his backup months ago!!!!!!! And even that drive was going out... I decided he needed a Raid 10 Array with a different hard drive manufacturers...

Just Joking!

I decided he needed Amazon S3... As at least the monthly bill will not be very high for all his data storage needs.

Hitachi drives seem good... Too bad WD bought Hitachi?

B.Thomas · Post by **B.Thomas** » Sun Nov 11, 2012 10:58 pm

I was fishing around on the Amazon site and they also host MySQL databases for deployment. http://aws.amazon.com/rds/mysql/

Does that mean we could run our MySQL database from the Amazon "Cloud"?

Justin Shafer · Post by **Justin Shafer** » Sun Dec 30, 2012 6:29 pm

I don't see why not.. You would want to store the images inside the database? Is it HIPAA Compliant? I know a way to get a bucket to be hipaa compliant, but a bucket may not be the same as a mySQL instance.

Q: Which relational database engines does Amazon RDS support?
Amazon RDS currently supports MySQL 5.1 and 5.5 (Community Edition) with InnoDB as the default database storage engine, Oracle Database 11gR2 and SQL Server 2008 R2.

caringdentists · Post by **caringdentists** » Fri Jan 04, 2013 11:27 am

Folks, I have had 2 server crashes!
Situation 1: I had 2 mirrored drives using BIOS from the motherboard (SATA drives). I had to break the mirror.
Situation 2: No hardware failure but after the Windows update, some bizarre unexplainable Windows OS got corrupted. (not from user error or virus). Had to reinstall windows.

I've used MS Synctoys, Norton Ghost 15. etc etc. But from my experience it is best to keep it really simple.
1. Create a batch file. " baclkup.bat"
2. Edit with notepad:

[quote]net stop mysql
ping 192.168.0.3
robocopy C:\XDR\XDRData\ m:\XDR\XDRData\ /s /v /r:1 /LOG:backup-log.txt /L /NS /NC /NFL /NDL /TEE
robocopy C:\mysql\data\replicate m:\mysql\data\replicate /s /v /r:1 /w:1 /eta /LOG+:backup-log.txt /L /NS /NC /NFL /NDL /TEE
robocopy C:\Opendentaldata m:\Opendentaldata /s /v /r:1 /w:1 /eta /LOG+:backup-log.txt /L /NS /NC /NFL /NDL /TEE
net start mysql[
notepad c:\backup-log.txt
/quote]

3. map a drive M: to a PC running OD and XDR (root directory C:\)
4. Schedule this batch file to run every night
This is the easiest way I have found because:
1. It automatically stops MySQL
2. The batch file can be copied to several machines without being altered. The only thing that changes is the individual PC's drive mapping. So you can do this infinitely. Scheduling the tasks sequentially from PC to PC.
3. Note that the program "robocopy" is an "internal command" in Microsoft! So no extra software is needed.
4. If there's a crash software, hardware whatever. I just point to the appropriate PC mysql, remap the Z: drive (for XDR) . Done.

Here's the log file to confirm:

-------------------------------------------------------------------------------
ROBOCOPY :: Robust File Copy for Windows
-------------------------------------------------------------------------------

Started : Fri Jan 04 14:01:41 2013

Source : C:\XDR\XDRData\
Dest : m:\XDR\XDRData\

Files : *.*

Options : *.* /V /NS /NC /NDL /NFL /L /TEE /S /COPY:DAT /R:1 /W:30

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Total Copied Skipped Mismatch FAILED Extras
Dirs : 3136 0 3136 0 0 3
Files : 34328 0 34328 0 0 8
Bytes : 8.021 g 0 8.021 g 0 0 12.3 k
Times : 0:01:02 0:00:00 0:00:00 0:01:02

Ended : Fri Jan 04 14:02:43 2013

-------------------------------------------------------------------------------
ROBOCOPY :: Robust File Copy for Windows
-------------------------------------------------------------------------------

Started : Fri Jan 04 14:02:44 2013

Source : C:\mysql\data\replicate\
Dest : m:\mysql\data\replicate\

Files : *.*

Options : *.* /V /NS /NC /NDL /NFL /L /TEE /S /COPY:DAT /ETA /R:1 /W:1

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Total Copied Skipped Mismatch FAILED Extras
Dirs : 1 0 1 0 0 0
Files : 652 0 652 0 0 0
Bytes : 128.05 m 0 128.05 m 0 0 0
Times : 0:00:00 0:00:00 0:00:00 0:00:00

Ended : Fri Jan 04 14:02:44 2013

-------------------------------------------------------------------------------
ROBOCOPY :: Robust File Copy for Windows
-------------------------------------------------------------------------------

Started : Fri Jan 04 14:02:44 2013

Source : C:\Opendentaldata\
Dest : m:\Opendentaldata\

Files : *.*

Options : *.* /V /NS /NC /NDL /NFL /L /TEE /S /COPY:DAT /ETA /R:1 /W:1

------------------------------------------------------------------------------

------------------------------------------------------------------------------

Total Copied Skipped Mismatch FAILED Extras
Dirs : 6745 26 6719 0 0 0
Files : 12302 0 12302 0 0 13
Bytes : 1.662 g 0 1.662 g 0 0 12.73 m
Times : 0:01:19 0:00:00 0:00:00 0:01:19

Ended : Fri Jan 04 14:04:04 2013

caringdentists · Post by **caringdentists** » Fri Jan 04, 2013 11:34 am

I've also used Microsoft" "backup and restore". and "Windows drive mirroring". Nothing is better than the batch file I've written. Because you can customize so many more things.

Justin Shafer · Post by **Justin Shafer** » Sun Jan 06, 2013 6:22 pm

Check out SyncBack Pro.. look at the options in Expert Mode..

I dare you. You might like it.

Robocopy is a great choice with batch files... Only one I like more is rsync and that is mainly for backing over WAN.

Raid or mirroring.. is NOT a backup.. its to protect the data sitting on the array in realtime. To protect against a single drive failure and then restoring your latest backup.

I implore you.. use Raid 1.

irfan · Post by **irfan** » Sat Feb 02, 2013 8:45 am

Does that work better than Crashplan? I like that Crashplan backs up to my office, an external drive, and my home computer.

irfan · Post by **irfan** » Sat Feb 02, 2013 8:48 am

with the robocopy... is it required to really stop the mysql? I dont stop for crashplan, and i think i remember reading that Jordan said it wasnt a big deal... so would it be possible to use that to copy the files every 15mintues... or does it not do incremental backups and only the full directory? I like the idea of it better since there is no exporting from crashplan required.

Post by **jsalmon** » Sat Feb 02, 2013 11:40 am

Stopping the MySQL service isn't required but highly recommended. There have been issues with tables being "corrupt" in the backup because the table was temporarily locked by MySQL because it was writing to it or something of that nature. To be 100% sure the backup will be a spitting image at the time it is taken, the service should be stopped. However, we do not have our service stop with our crash plan pro at work and I've loaded up several databases from it and they've been fine. The story is not the same with other backup programs because every single one of them seems to do things differently.

Justin Shafer · Post by **Justin Shafer** » Fri Apr 12, 2013 3:42 am

jsalmon wrote:Stopping the MySQL service isn't required but highly recommended. There have been issues with tables being "corrupt" in the backup because the table was temporarily locked by MySQL because it was writing to it or something of that nature. To be 100% sure the backup will be a spitting image at the time it is taken, the service should be stopped. However, we do not have our service stop with our crash plan pro at work and I've loaded up several databases from it and they've been fine. The story is not the same with other backup programs because every single one of them seems to do things differently.

That reminds me of school. We had a teacher that warned us of magical backup programs that just seemed to more or less be magic. Being able to backup a database while users are still logged in, data could still be in memory etc. Of course this was a the dawn of Volume Shadow Copy.. and.. we were told more or less.. not to trust if when it came to our but's being on the line. If we can.. at anytime..stop the database, back it up, and then start it. This was with "any database".. type conversation.. not just mysql.

I think he gave good advice. Be on the "safe side". mysqlhotcopy?? or the mysql dbbackup program.. I forget the name.. at least it's supported by mysql.. I assume it locks all the tables down and does a dump? While locked, user requests are.. ?????

Jay · Post by **Jay** » Fri Apr 12, 2013 4:29 pm

I agree, Justin.

Some examples of how others have done this for educational purposes only.

http://dev.mysql.com/doc/refman/5.1/en/mysqldump.html

stjames70 · Post by **stjames70** » Sun Apr 14, 2013 2:03 pm

12 April 2013:

Drive #3 of our Macpro suddenly disappears from our four drive array at 3:00 pm in the afternoon.
Bad moomoo. That drive contains our XpPro server so all of our workstations lose contact with the server.
I reboot our server and drive #3 shows up again. I clone the VM of our server to drive #2 of our array. 3:30 by the time that is done.
I reboot the XP pro VM at 3:45 and no loss of data is noted. We are up and running within 45 minutes of possible hard drive failure.
The office carries on until closing time. In the interim between 3:45 pm and 7:30 pm I have one of my partners run to MicroCenter and acquire a 3TB WD Red drive. At 7:45, we shut down the XP Pro VM, take out the old drive #3, and insert the newly acquired drive on bay #3. I reclone the XP Pro machine back to the new drive and reset all the settings of Carbon Copy Cloner which has been our trusty bit copier for our main drives.

At 8:30 pm I go home, smug and happy that our backup and disaster recovery protocols have served us well. RAID 1+0 array is next in the budget. RAID card and testing in progress -- just waiting for budget to acquire 2TB worth of SSD storage in a 1+0 array. Even if our server motherboard had completely failed, the internal drives of our server are replicated on a nightly basis. So at most, loss of one day's worth of data. Bottom line= 45 minutes downtime, 5.5 hours to reconstitute all of our back up policies and hardware. VMs rule. The clone contained all the network settings of the original, so no wasted time tweaking with those parameters.

tacomadental · Post by **tacomadental** » Thu Apr 26, 2018 9:51 am

I have my server in Raid 1, then remove one HD from down server, replace down server with my temp server, attach external HD docking station place that HD with the most current info in external docking station and boot up direct opendental and xdr to the external drive till I can have Fry's build me a new rig.

aliallam22 · Post by **aliallam22** » Sun Jul 15, 2018 8:39 pm

We used iDrive for backup -- everyday at 1am, it automatically backs-up OpenDentImages and the mysql database. One time when the server went down, we just restored the latest backup onto one of the workstations, called OD, and they helped us download mysql. Then we just set it up as a temporary server.

enamelrod · Post by **enamelrod** » Wed Jul 18, 2018 7:42 am

1. separated apteryx and open dental and my main server onto 3 different machines. But each machine has a working copy of apteryx and open dental executable files.

2. all data is stored on a nas box synology 5 disk in raid

3. all 3 data sets are backed up onto the actual servers every 15 minutes

4. duplicate backed up into the sky every 30 minutes from the main server with 2 different programs.

5. my keys have a 128 gb key drive that i back up every day im actually in the office and If I remember

if OD goes down its a simple point to new server and done
if my main server goes out we have other issues but we have a computer to run apteryx and opendental and point back to NAS
if apteryx goes out I install and point back to the nas
If NAS goes out. I point all data links back to the individual servers

if the building burns down. i have a copy at home
if my home is destroyed in a tornado. the underwater Microsoft data center hopefully will survive the nuclear holocaust but i wont have any patients that I can take xrays on so I only need my copy of OPEN DENTAL on my keys